An analysis of the nature of the gene action for seven characters influencing productivity and wide adaptation in Linum usitatissimum L. was undertaken in this investigation over three seasons based on diallel and fractional diallel crosses, among a set of ten genetically diverse parents. Estimation of combining ability and the nature of geno-type-environment interactions for the above characters was also examined.A comparison of the means of the parents and hybrids over three environments showed positive heterosis for height, number of tillers, fruit-bearing branches and capsules per plant while negative heterosis was observed for flowering time. The remaining two characters viz., height at branching and seeds per five capsules showed little differences due to mutual cancellation of the components of heterosis.A serial analysis of combining ability over three seasons using GRIFFING's model (1956) has revealed significant sea-sonal effects on the estimates of combining ability effects. Additive gene action was found to be predominant in most of the characters although non-additive variation due to s.c.a. was equally prominent for fruit-bearing branches and capsules per plant. On the basis of g.c.a. for all the characters, M. 10 and N. 55, both Peninsular types were the best followed by N.P. (R.R.) 45 and Mohaba Local.In the fractional diallel analysis the relative estimates g.c.a. and s.c.a. were similar for s=9 and 7. However, there was overestimation in favour of s.c.a. when s was reduced to 3. The serial analysis of fractional diallel had confirmed the results of full diallel. Significant genotype x environment interactions for flowering time, height, height at branching, fruit-bearing branches and capsules per plant were observed in this analysis also. The results have indicated that reduction of the diallel to a fraction with s=n/2 would vitiate the order and precision of estimates of combining ability.The analysis of genotype x environment indicated that the pattern of interactions among the parents of different geographical groups was not similar but varied with the character under study. Considerable interactions were observed for capsules per plant, fruit-bearing branches and flowering time in that order respectively. A comparison of the parental and hybrid populations for the degree of interaction for all the characters had revealed that heterozygosis might not at all be directly related to the degree of homeostasis. Actually, heterozygotes were found to show more interactions than homozygotes for flowering time, height at branching and tiller number and lower interactions for the other four characters. Significant seasonal effect on the degree of heterosis for vegetative as well as reproductive characters was observed in a majority of F1's with variable degree of dominance.